?
Academy of Managemenit Journal
1999, Vol. 43, No. 3, 239-259.
MULTIMARKET CONTACT, ECONOMIES OF SCOPE, AND
FIRM PERFORMANCE
JAVIER GIMENO
Texas A&M University
CAROLYN Y. WOO
University of Notre Dame
We integrate the efficiency and competitive effects of product-marketscope choice into
a comprehensive model of economic performance and empirically test the model in the
context of the U.S. airline industry. Efficiency is influenced by a firm's scope economies, but the intensity of rivalry is determined by multimarket contact with rivals and
their scope economies. The confluence of strong scope economies with multimarket
contact results in superior economic performance. However, strong scope economies
may not result in superior performance if rivals can obtain similar economies in
nonoverlapping markets.
Few strategic decisions are as important for a
firm as its choice of scope-the set of products and
markets in which it will compete. Prior research
has investigated how the presence of economies of
scope, synergies, or relationships between product
markets (Ansoff, 1965; Panzar & Willig, 1981; Porter, 1985) determines the efficiency associated with
product-market choices. The literature suggests
that efficiency is greater and performance higher
when a firm's operations in multiple products or
markets share common resources or value-chain
activities (Brush, 1996; Panzar & Willig, 1981; Penrose, 1959; Porter, 1985). The choice of productmarket scope also determines the set of competitive
relationships with other firms. As firms expand
into new markets, they may encounter new competitors as well as some competitors from their
original markets that have undertaken similar expansions (Porter, 1985). Hence, firms often attempt
to exploit scope economies in the context of other
competitors' pursuing similar product-market extensions, a situation that may lead to multimarket
competition. Generally, economies of scope and
multimarket competition are likely to occur concurrently. Surprisingly, prior research has examined these constructs independently, ignoring their
possible interdependence and interaction in determining economic performance.
The contribution of this research is the theoretical and empirical integration of the efficiency and
competitive effects of product-market scope on economic performance. Even if economies of scope
make firms more efficient, those economies may
not result in superior performance if rivals are able
to draw on similar economies and are motivated to
compete intensely. Hence, we investigated the
competitive effect of rivals' scope economies and
their multimarket contact with a focal firm. The
integrative approach also allowed us to explore the
interaction between multimarket contact and scope
economies, as managers may more readily recognize their interdependence with multimarket rivals
when scope economies are present. The efficiency
and competitive dimensions were integrated into
an internally consistent model of economic performance. To clarify the causal structure of the model,
we empirically specified two mediating dependent
constructs, efficiency and intensity of rivalry, reflecting the efficiency and competitive dimensions,
as well as the ultimate dependent construct representing economic performance, profitability. Hypotheses were tested in the context of the U.S.
airline industry.
We thank Michael Hitt, Robert Hoskisson, Rajan Varadarajan, and seminar participants at the University of
Texas at Austin for valuable suggestions. We also gratefully acknowledge very helpful comments on manuscript
drafts from Angelo De Nisi and the reviewers of the
Academy of Management Journal. The first author acknowledges the financial support of the Spanish Ministry of Science and the Purdue Research Foundation.
BACKGROUND
Since this article integrates theories from different research streams and disciplines, it is useful to
define some key concepts as they are used here.
Our research deals with firms that are present in
multiple markets. In agreement with economic the239
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
240
Academy of Management Journal
ory, we use the term market to represent the aggregate demand for a product or service in a given
geographical area, with market boundaries determined by low cross-elasticity of demand. In practice, markets can be outlined as different products
or services within an industry, as geographical demand for a given product or service, or even as
different industries.1 The presence of a multimarket firm-that
is, a firm with a multimarket
scope-in a specific market will be termed a market-unit.
Economies of Scope
The concept of economies of scope describes the
cost savings that result from the activities of a firm
in multiple markets (Panzar & Willig, 1981). Economies of scope are gained when the costs to a single
firm of producing a given level of output for each of
several markets are lower than the summed costs of
separate firms each producing at the given output
level for a single market (Bailey & Friedlaender,
1982). Economies of scope are commonly stated as
a condition of cost subadditivity in a joint cost
function for multiple markets C(y1, Y2), with:
C(y1, Y2) < C(y1, 0) + C(O, Y2),
where C(y1, Y2) is the joint cost of producing Yi
units of output for market 1 and Y2 units for market 2.
Economies of scope generally arise from the sharing or joint utilization of inputs of production (resources, in our terminology), when the shared inputs are quasi-public-that
is, they are developed
or acquired for use in one market and are available
freely or at reduced additional cost to other marketunits (Bailey & Friedlaender, 1982; Panzar & Willig,
1981). Therefore, resource-sharing opportunities
across markets are an antecedent of economies of
scope.2 If transaction costs prevent the existence of
an efficient market in the shared resources, econo1 Arguably, the definition of a market is partly dependent on the level of aggregation, since many industries
actually entail multiple products, and national markets
can often be divided into more meaningful regional or
local markets. Economics researchers have developed
econometric tests for the determination of market boundaries in terms of cross-elasticity of demand (Scheffman &
Spiller, 1996).
2 Bailey and Friedlaender (1982) suggested some common specific conditions for economies of scope, including (1) resources jointly produce multiple by-products,
(2) fixed indivisible resources can be used for multiple
markets, (3) economies of networking exist, (4) a resource
can be reused by more than one product, and (5) markets
share intangible resources.
June
mies of scope may lead to the emergence of multimarket firms (Penrose, 1959; Teece, 1980).3
Although economies of scope have often been
evaluated at the firm level of analysis, a recent
trend is to analyze them at a more disaggregate
(market-unit) level (Brush, 1996; Davis & Thomas,
1993). This trend provides a finer-grained representation of scope economies, since all market-units
within a firm are not likely to benefit equally from
resource-sharing opportunities with other marketunits. For instance, if a firm has two market-units
that share the same resources and a third that does
not share resources with the others, the market-unit
level of analysis would accurately describe the intrafirm differences in scope economies among
these market-units. At the market-unit level, the
condition of economies of scope can be written as
[C(y1, Y2) - C(O, Y2)]/Yl < C(y1, O)I/Y1
That is, the incremental average unit cost of producing output 1 when the firm is also producing
output 2 is lower than the average unit cost of
producing output 1 on its own. More generally,
economies of scope imply lower average costs for a
focal market-unit if the firm also has other marketunits with resource-sharing opportunities with the
focal market-unit.
Multimarket Contact
Multimarket contact occurs when firms encounter the same rivals in multiple markets. When firms
compete with each other in several markets-a phenomenon also known as multimarket (or multipoint) competition (Karnani & Wernerfelt, 1985;
Porter, 1980, 1985)-their
competitive behavior
may differ from that of single-point rivals. Multimarket competition may result in the reduction of
the competitive intensity among rivals, an outcome
known as mutual forbearance (Edwards, 1955). A
firm that meets a rival in multiple markets can
respond to an attack not only in the attacked market, but also in other markets in which both firms
compete. The competitive moves of multipoint
competitors may therefore be linked across markets; this condition is known as extended interde-
3 In addition to economies of scope (cost subadditivity), multimarket firms may also benefit from revenue
superadditivity. Revenue superadditivity occurs when
the volume of sales by the firm in one market increases
the demand for the firm's products in another market.
Revenue superadditivity may exist in the demands of
complementary products (such as razors and blades) or
in markets with cross-market demand externalities.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1999
Gimeno and Woo
pendence (Areeda & Turner, 1979). As firms recognize their rivals' ability to retaliate in multiple
markets, they develop expectations about crossmarket retaliation (Feinberg, 1984) that will reduce
their motivation to act aggressively (Chen, 1996).
Since the retaliatory power is reciprocal, the forbearance is mutual. In addition to mutual deterrence, multimarket contact may also increase a
firm's familiarity with the strategies of its rivals,
which may also facilitate the tacit coordination and
mutual understanding necessary to successfully reduce rivalry (Baum & Korn, 1996; Scott, 1993).
Empirical research on mutual forbearance has
shown inconclusive results in the past, although
recent studies using longitudinal research designs
have generally provided consistent support for the
mutual forbearance hypothesis (see Jayachandran,
Gimeno, and Varadarajan [1999] for a review).
However, the lack of consideration of scope economies in empirical multimarket contact research
casts doubt over some empirical findings since, as
Montgomery pointed out, "Whether this [the performance effect of multimarket contact] was due to
natural scope economies, anti-competitive behavior, or both was not clear" (1994: 169). In the theory
and analysis presented below, we seek to clarify
this important question.
THEORY AND HYPOTHESES
In the theory presented here, we first hypothesize
an association between economies of scope and
multimarket contact and then explore the performance effects of these two constructs. Although the
ultimate dependent variable of interest is economic
performance (profitability), we begin by explicitly
examining the effects of these constructs on two
mediating variables, efficiency and intensity of rivalry. Efficiency and intensity of rivalry are two
important predictors of profitability (McWilliams &
Smart, 1993; Schmalensee, 1987), and they are the
theoretically expected mechanisms by which scope
economies and multimarket contact influence profitability.
In order to tease out the effects of scope economies and multimarket contact, we defined the level
of analysis at the market-unit rather than the aggregate firm level. This is a very important decision,
since firm-level aggregation of effects may mask
significant intrafirm differences among marketunits. Different market-units within a firm may
benefit differently from scope economies, depending on their ability to share resources with the
firm's other market-units (Brush, 1996; Davis &
Thomas, 1993). Competitive effects are also likely
to differ across market-units, since firms are likely
241
to encounter different combinations of competitors
in each market. In this article, the term focal market-unit refers to the market-unit being analyzed in
a given observation, and the term focal-market
rivals identifies the rivals that a focal firm faces in
a focal market. Figure 1 represents the main relationships studied here.
Association between Economies of Scope and
Multimarket Contact
Theory and evidence suggest that firms' expansion paths are partly determined by the incentive to
find additional uses for existing resources that are
not fully used in existing markets (Montgomery &
Hariharan, 1991; Penrose, 1959; Teece, 1980). Such
activities can lead to scope economies from the
sharing of resources across markets. These market
expansion opportunities may not be totally specific
to each firm. Firms in the same focal market may
have developed similar resources for serving that
market and may therefore consider similar market
expansion options. To the extent that market expansion options based on resource-sharing opportunities are sufficiently visible, competitors are
likely to independently perceive the gains from the
same market expansions, so multimarket contact
results. As Porter argued in the context of corporate
diversification, "While multipoint competitors and
interrelationships do not necessarily occur together, they often do because both tangible and
intangible interrelationships lead firms to follow
parallel diversification paths" (1985: 354). Therefore, although multimarket contact can also occur
in the absence of economies of scope, the likelihood of meeting focal-market rivals in other markets that offer strong resource-sharing opportunities with the focal market should be greater than it
will be in those with weak resource-sharing opportunities.
Hypothesis 1. Multimarket contact between a
firm and its focal-market rivals is more likely
to occur in markets that present strong resource-sharing opportunities with the focal
market than in those that present weak resource-sharing opportunities.
Effects on Market-Unit Efficiency
The following hypothesis addresses the impact
that resource-sharing opportunities with other market-units have on the efficiency of a focal marketunit. Efficiency, defined as the ability of a firm to
produce a given level of output with fewer inputs
and resources (or greater output with a given level
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
242
June
Academy of Management Journal
FIGURE 1
Model of the Effects of Multimarket Contact and Economies of Scope on Market-Unit Performance
........................................................................
Conditions in other markets (n)
served
by firm i
.........-
Effects for focal firm i's market-unit
in focal market m
Strong resource-sharing
opportunities
(between focal market m
and market n)
~ ~~ ~ ~ ~
.~
-------------Economies
of scope
~~~~~~--------------
Cost efficiency
(fir i, marketm)
L
(firm i, marketi marketm)
_____________________________________
.
. \
\*
*.
:
Multimarket contact
(firm i's rivals in focal
m are also present
market
market
in market n)
*
:
* _______________________________-forbearance
.
........................................
~~~~~~~~~~~~~~
.
~~~~~~~~~~~Profitability
.
~
*
I
~~~~~~----------------------------/
Mutual
~~~~~~~~~~~forbearance
Intensity of rivalry
(firm i, m arket m )
/--------------------
Strong resource-sharing .
opportunities
(between focal market m
and market n')
I
I
.
.
.
...
I
Conditions in other markets (n')
served by focal-market rivals
but not by focal firm i
---\--
-
- -
-/-
-
of resources), is an important predictor of competitive advantage within a market or industry (Demsetz, 1973; Peteraf, 1993; Schmalensee, 1987).
Higher efficiency has been cited as one of the most
salient potential benefits of sharing resources
across the value chains of different market-units
(Bailey & Friedlaender, 1982; Brush, 1996; Porter,
1985). The joint utilization of resources improves
efficiency by increasing the rate of output that a
firm obtains from existing resources. This is especially so if the underlying shared resources represent fixed costs or if the costs of the resources
increase at a less than proportional rate relative to
their utilization.
Operationally defined at the market-unit level of
analysis, the concept of economies of scope suggests that the efficiency achieved by a focal marketunit is greater when it is able to share resources
with other market-units of its firm. Therefore, we
expected a positive efficiency spillover across market-units within a firm when the market-units have
strong resource-sharing opportunities.
Hypothesis 2. The efficiency of a focal marketunit is higher if the focal firm is present in
other markets that have strong resource-sharing opportunities with the focal market.
Effect on Market-Unit Intensity of Rivalry
We now shift focus from efficiency to intensity of
rivalry as the dependent construct. The intensity of
rivalry experienced by a focal market-unit is determined by the competitive interaction with focalmarket rivals (Porter, 1980), and it is therefore influenced by the competitive behavior of those
rivals. The product-market scope choices of focalmarket rivals may influence the intensity of rivalry
experienced by the focal market-unit if these
choices lead to multimarket contact between the
focal market-unit and its focal-market rivals. The
mutual forbearance hypothesis suggests that the
threat of cross-market retaliation among competitors that have substantial multimarket contact reduces the intensity of their rivalry. Past research
has yielded evidence that multimarket contact is
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1999
Gimeno and Woo
associated with constraint of rivalry, as reflected by
higher prices (Evans & Kessides, 1994; Gimeno,
1999; Gimeno & Woo, 1996a), lower frequency and
speed of competitive moves (Young, Smith, &
Grimm, 1997), higher conjectural variations (Parker
& R6ller, 1997), and lower likelihood of market
entry (Baum & Korn, 1996) and exit (Barnett, 1993;
Baum & Korn, 1996; Boeker, Goodstein, Stephan, &
Murmann, 1997). Given the substantial amount of
prior testing of this relationship that has occurred,
the next hypothesis, replicating prior findings, was
included for theoretical completeness of the model
rather than for its original contribution.
Hypothesis 3a. The intensity of rivalry experienced by a focal market-unit is negatively associated with the extent of multimarket contact
with its focal-market rivals.
Existing research on multimarket contact is limited by the established practice of aggregating all
contacts into a summary measure. By doing so,
researchers have not acknowledged that different
types of contacts may have different collusive effects. In the next hypothesis, we differentiate between contacts in markets that have strong resource-sharing opportunities with a focal market
and those that do not, in the expectation that these
contacts may have different effects on the intensity
of rivalry.
A necessary condition for mutual forbearance is
that firms recognize their extended interdependence. Such recognition requires firms to actively
consider the possibility of cross-market retaliation.
Discussing multipoint competition across diversified firms, Porter suggested that "strategy towards a
multipoint competitor is affected by whether or not
the competitor perceives the connections among
industries" (1985: 359; emphasis in original) and
that "relatedness increases the likelihood that a
competitor will perceive the linkages among businesses" (1985: 360). If managers are attempting to
exploit economies of scope among market-units,
decisions concerning these market-units will probably be more coordinated in the organization's decision-making process. This coordinated decision
making facilitates the perception of extended interdependence with multipoint competitors present
in those markets. Even if the market-units are managed by different organizational units, the presence
of strong resource-sharing opportunities among
market-units may lead to the formation of coordinating roles and mechanisms (Hill, Hitt, & Hoskisson, 1992), which would permit the recognition of
interdependence with multipoint rivals. On the
other hand, failure to perceive extended interdependence or lack of effective coordination would
243
likely reduce the ability to synchronize response to
rivals across units and thus undercut the incentives
for forbearance (Collis & Montgomery, 1997: 166). It
is therefore reasonable to expect that unless the
opportunities for resource sharing among units are
strong and can be easily recognized, the coordination necessary to induce forbearance will largely be
absent.
The gains from tacit collusion among multipoint
competitors are also more limited in markets that
do not benefit from economies of scope with other
markets. This is because, in the absence of economies of scope, multimarket competitors do not
have an efficiency advantage over single-point incumbents or potential entrants. Thus, an attempt by
multimarket incumbents to collude (for instance,
by raising their prices over their costs) may be
self-defeating, since single-point incumbents and
potential entrants would act to bring the market
back to competitive equilibrium. In contrast, when
multimarket competitors can draw on the benefits
of economies of scope, they have a cost advantage
over single-point incumbents or potential entrants.
They can raise prices to a level that, although above
the costs of the multimarket incumbents, does not
generate output expansion by single-point incumbents and potential entrants. Thus, the opportunity
to tacitly reduce rivalry may be greater when multimarket competitors meet in multiple markets
characterized by strong resource-sharing opportunities.
These rationales therefore suggest that the effects
of multimarket contacts are not homogeneous: multiple contacts in markets with strong resource-sharing opportunities with the focal market are more
likely to reduce the intensity of rivalry experienced
by the focal market-unit.
Hypothesis 3b. The reduction of rivalry from
multimarket contacts is greater for contacts in
markets that present strong resource-sharing
opportunities with the focal market.
Intensity of rivalry may also be influenced by the
focal-market rivals' presence in markets in which
the focal firm is not present-that is, when multimarket contact does not occur. Without the competitive restraint arising from multimarket contact,
focal-market rivals are likely to exhibit competitive
behavior in accordance with their competitive capability (Chen, 1996). We hypothesize that the intensity of rivalry will increase if the nonoverlapping product-market scope of rivals provides them
with strong resource-sharing opportunities for their
units in the focal market. In this case, the focalmarket rivals enjoy a source of advantage and a
potential for greater efficiency not available to the
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
244
Academy of Management Journal
focal market-unit. The source of this advantage also
lies outside the range of retaliation by the focal
market-unit. In that situation, focal-market rivals
are likely to deploy their advantage to obtain a
larger share of the focal market, thus increasing the
intensity of rivalry experienced by the focal market-unit. As a result, the intensity of rivalry for the
focal market-unit will be higher if focal-market
rivals enjoy efficiencies gained from resource-sharing opportunities with other markets in which the
focal market-unit is not present.
Hypothesis 4. The intensity of rivalry experienced by a focal market-unit is higher if focalmarket rivals are located in markets that
present strong resource-sharing opportunities
and in which the focal firm is not present.
Effects on Market-Unit Profitability
We now examine the impact of multimarket contact and economies of scope on the variable of
ultimate interest, profitability. Since profitability,
or economic performance, is a combined result of
efficiency and intensity of rivalry (Schmalensee,
1987), the hypotheses below extend the foci of the
two prior sets of hypotheses to their ultimate implications. By simultaneously including economies
of scope and multimarket contact in our model of
profitability, we avoid the possible omitted variable bias implied in Montgomery's (1994) quote.
Explanations of profitability differences in strategic management and industrial economics have
emphasized two generic determinants of profitability (McWilliams & Smart, 1993; Schmalensee,
1987). Profitability may be due to "quasi-rents" associated with firm-specific resources and capabilities that make firms more efficient (Cool, Dierickx,
& Jemison, 1989; Demsetz, 1973; Peteraf, 1993).
Profitability may also be due to market power attained when external conditions, such as industry
structure, determine a low level of competition
from rivals, potential entrants, buyers, and suppliers (Porter, 1980; Scherer & Ross, 1990). Efficiency
and market power explanations of profitability
have often been presented as conflicting views, but
they are clearly not mutually exclusive and are
treated here as complementary influences.
The profitability of a market-unit is likely to be a
function of its cost efficiency, since market-units
with lower costs will also obtain larger margins,
everything else being constant. Accordingly, factors that increase market-unit efficiency should
also increase market-unit profitability. Hence,
repeating and extending the arguments used for
Hypothesis 2, we predicted that strong resource-
June
sharing opportunities would increase market-unit
profitability.
Hypothesis 5. The profitability of a focal market-unit is higher if the focal firm is present in
other markets that have strong resource-sharing opportunities with the focal market.
The profitability of a market-unit is also likely to
be a function of its competitive environment and,
particularly, of the intensity of rivalry generated by
focal-market rivals. If rivalry is intense, focal market-units will lose sales to those rivals undertaking
competitive actions. Focal market-units may reduce the loss in revenues by stepping up their own
competitive activity (by, for example, reducing
their price or increasing promotional and differentiation efforts), but by doing so they may sacrifice
margins. In either case, profitability is hurt by intense rivalry, and factors that reduce the intensity
of the rivalry experienced by a market-unit should
also increase the market-unit's profitability. Repeating and extending the arguments used for Hypotheses 3a, 3b, and 4, we theorize that market-unit
profitability will be positively associated with multimarket contact, particularly when contacts occur
in markets with strong resource-sharing opportunities, and that it will be negatively associated with
facing focal-market rivals that have strong resourcesharing opportunities in nonoverlapping markets.
Hypothesis 6a. The profitability of a focal market-unit is positively associated with the extent
of multimarket contact with its focal-market
rivals.
Hypothesis 6b. The increase in profitability
from multimarket contacts is greater for contacts in markets that present strong resourcesharing opportunities with the focal market.
Hypothesis 7. The profitability of a focal market-unit is lower if focal-market rivals are located in markets that present strong resourcesharing opportunities with the focal market but
in which the focal firm is not present.
METHODS
Sample
Although the constructs of economies of scope
and multimarket contact can be applied to multiple
contexts of multimarket scope (multiproduct industries, corporate diversification, and domestic or
international geographical diversification), here we
tested the hypotheses in the context of the multiple
markets within the U.S. airline industry. Data on
the scheduled air transportation of passengers were
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1999
Gimeno and Woo
used. The industry is composed of multiple citypair markets (customers demanding air transportation between two specific cities), with most airlines
present in several of these markets simultaneously.
The city-pair market offers a convenient definition
of a market, since there is very little or no crosselasticity of demand across city-pair markets. The
presence of an airline in a city-pair market is referred to here as an airline route. The airline industry is the single or dominant business for most
firms in the industry, which reduces any bias from
the omitted effect of multimarket contact outside
the industry.
This industry is ideal for the study of economies
of scope and multimarket competition for various
theoretical and empirical reasons. In terms of economies of scope, airlines engage in substantial sharing of resources across airline routes when multiple
routes share airport facilities (for instance, ticketing
and baggage-handling facilities, maintenance and
hangar facilities, and proprietary airport investments) or common segments of service within a
hub-and-spoke network. Airlines also experience
substantial multimarket contact (Baum & Korn,
1996; Chen, 1996; Evans & Kessides, 1994; Gimeno
& Woo, 1996a; Smith & Wilson, 1995). The few
players in the industry engage in repeated competitive interaction in multiple markets, which means
that firms mutually learn about the expected responses of their rivals. Anecdotal evidence of
cross-market retaliations (Nomani, 1990; O'Brian,
1994) reflects the extended interdependence among
airlines.
A panel data sample describing the domestic
scheduled passenger activities of U.S. airlines for
the fourth quarters of 1984 through 1988 was obtained from three U.S. Department of Transportation quarterly databases: the origin and destination
(O & D) survey (a 10 percent sample of all the
tickets sold in the United States), the service segment database, and the form 41 reports. The sample
period represents a critical window in the evolution of the U.S. airline industry and incorporates
phases of fierce rivalry as well as times of deescalation (Gimeno & Woo, 1996a). We focused on both
nonstop and one-stop services for city-pairs, since
these services are often considered as substitutes
for each other in the airline competition literature.
We only considered city-pair markets in which
both end-cities were at least small hubs, according
to the Federal Aviation Administration (FAA) classification (that is, enplanements in each airport
were at least 0.05 percent of the total yearly U.S.
enplanements). We also eliminated markets between cities less than 100 miles apart (to avoid the
effect of substitution by ground transportation) and
245
those with average daily traffic of fewer than ten
passengers. Since research focused on competitive
contexts, we also eliminated observations of monopoly markets. The application of these restrictions yielded 3,008 valid city-pair markets.
The unit of analysis (the focal market-unit) was
defined as the airline route. The sample included
14,122 airline routes. For each, an observation was
created for each time period in which the airline
route was active. The subscripts i, m, and t respectively identified the airline, city-pair market, and
time period of an observation. An airline was considered to be an incumbent in a market if it met at
least one of these two conditions: (1) it had at least
a 5 percent share of the market or (2) it carried at
least ten passengers a day.4 A potential entrant was
defined as a nonincumbent with operations at both
end-cities of the city-pair market.5 Airline-route
passenger figures did not include those passengers
flying with tickets that combined multiple airlines
in a single trip (interline tickets). However, interline tickets were counted as part of the total market
passenger figures used for calculating market
shares and market structure variables. We also
eliminated from the sample airlines for which the
revenues obtained in the sampled markets differed
by more than 30 percent from their published aggregated revenues (these airlines were included for
calculations of market shares and of market structure and multimarket contact variables). These discrepancies could be the result of undersampling in
the 0 & D survey or of the airlines' substantial
presence in other markets not included in the sample (such as small nonhub cities). Our final data set
included 28 airlines and 44,493 observations.6
This definition eliminated cases in which passengers
flew an airline route through connections unintended by
a firm but kept in the sample small competitors that
targeted niches of demand in high-density markets.
5 Research in the airline industry has well-developed
protocols for the identification of potential entrants to
markets, thanks to the accumulated research on contestability (Berry, 1989; Brueckner, Dyer, & Spiller, 1992).
Our definition agrees with that used by Brueckner and
colleagues. Berry supported this definition by finding
that the odds of entry for a potential entrant already
established at both points was more than 18 times larger
than that for a firm only established in one city and that
it was 77 times larger than that for a firm not established
in either city.
6 Grouped by their FAA classifications,
the 28 sampled airlines were (1) majors: American, Continental,
Delta, Eastern, Northwest, PanAm, Piedmont, Republic,
TWA, United, USAir, and Western; (2) nationals: Air
California, Aloha, America West, Braniff, Jet America,
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
246
Academy of Management Journal
Dependent Variables
Efficiency. Efficiency was measured as cost per
revenue-passenger-mile, which was the cost of flying a paying passenger in a given airline route
divided by the distance between the endpoint cities, stated in cents per mile. Most of the costs in the
airline industry (with the exception of food costs,
which are about 3 percent of total costs) are incurred in making capacity (seats) available, and
they are incurred regardless of whether a seat is
filled or empty (Bailey, Graham, & Kaplan, 1985:
49). The efficiency of an airline route can therefore
be increased in two ways: by reducing the cost of
offering available seats and by achieving high utilization of those seats. Airlines do not report actual
costs per revenue-passenger-mile for each airline
route, but they can be constructed as a function of
the load factor (the percentage of filled seats) and
the cost per available-seat-mile. The cost per revenue-passenger-mile for each airline route equals
the product of the cost per available-seat-mile and
the ratio of available-seat-miles to revenue-passenger-miles, which is the inverse of the load factor.
The formula is
Cost per revenue-passenger-mileimt
- cost per available-seat-mileimt
June
Intensity of rivalry. Although rivalry entails a
pattern of competitive actions and reactions (Chen,
1996), its outcome is commonly reflected in decreased prices for the services provided by a firm.
This is particularly true for the airline industry, in
which price competition is the main dimension of
competition. Thus, we used a measure of price
known in the industry as yield to capture lack of
rivalry. Yield was defined as revenue per revenuepassenger-mile, or the average price paid by customers in an airline route divided by the distance
between the endpoint cities, stated in cents per
mile. Higher yields reflect less intense rivalry.
Profitability. Profitability represents the ability
of a firm to obtain revenues above costs. Unfortunately, airlines only report profitability at the airline level. To obtain a measure at the airline-route
level, we used -the variables yieldj-t (price per revenue-passenger-mile) and cost per revenue-passenger-mileimt to construct the Lernerindex. The latter,
a popular measure of economic performance used
in industrial economics, is defined as the price-cost
margin divided by the price. We calculated the
Lerner index for each airline route as:
Lerner indexi,t
yieldimt - cost per revenue-passenger-mileimt
yieldimt
available-seat -mileimnt
X 100.
revenue-passenger-mileImt
=
cost per available-seat-mileimt
1
X load factor,mt
The airline-route load factor is calculated from
the load factors of the segments used by passengers
traveling the route. We estimated airline-route cost
per available-seat-mile from firm-level and routelevel data following the procedure developed by
Brander and Zhang (1990).7
.The Lerner index is equivalent to the popular
return on sales (ROS) performance ratio, which is
defined as operating margins divided by sales. We
assessed the validity of the Lerner index by comparing firm-level ROS to a firm-level aggregation of
the Lerner index across airline routes. The salesweighted average of the Lerner index of a firm's
airline routes demonstrated a Pearson correlation
of .90 with firm-level ROS, suggesting substantial
convergent validity.
Independent Variables
Midway, Muse, New York Air, Pacific Southwest, Southwest, People Express, Southwest, and World; and (3)
large regionals: Air Atlanta, Florida Express, and Midwest Express. We used 21 other carriers to create citypair market variables but eliminated those carriers from
the final sample because the markets sampled here represented less than 70 percent of their overall revenues.
7 Brander and Zhang (1990) extrapolated unit costs for
an airline route from firm-level data. The costs per available-seat-mile are primarily determined by both firmlevel factors (such as labor costs or overall efficiency) and
operational factors. Operational costs per available-seatmile tend to decrease with market distance, since more
For each observation of a focal firm in a focal
market, the independent variables pertained to the
efficient planes can be deployed over longer distances.
The firm-level cost per available-seat-mile thus reflects
firm-level efficiencies as well as the "average flight
length" of the firm. Following Brander and Zhang (1990),
we calculated airline-route cost per available-seat-mile as
a distance-adjusted function of firm-level cost per available-seat-mile, as follows: cost per available-seat-mileimtcost per available-seat-mileit x (average flight lengthitl
distancem)l1/2.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
247
Gimnenoand Woo
1999
details of the definition of these variables are presented in Appendix A.
Focal firm's resource-sharing
opportunities.
The set SR[m] of markets that offer strong resourcesharing opportunities with a focal market was defined as those markets that share an end-city (origin
or destination) with the focal market. Airline routes
that share an origin or destination have the highest
opportunity for scope economies, since they can
share important ground resources such as gates,
maintenance and baggage-handling facilities, and
airport facilities, as well as common flight segments within a hub-and-spoke network. These
sharing opportunities are not available to airline
routes that do not share an origin or a destination.
To evaluate the effects of resource-sharing opportunities, we divided the set of markets served by a
focal firm (except the focal market) into two subsets: those with strong resource-sharing opportunities with the focal market (subset InSR[m]) and
those with weak resource-sharing opportunities
(subset InSR*[m]). The numbers of markets in
other (nonfocal) markets served by the focal firm
and its focal-market rivals. The variables described
two dimensions of these nonfocal markets: the
strength of resource-sharing opportunities with the
focal market, and the presence of the focal firm, the
focal-market rivals, or both, in these markets. The
Venn diagram in Figure 2 reflects the intuition behind our definition of the independent variables.
Three sets of nonfocal markets and their respective
complementary sets can be defined from the universe of markets. First, markets can be classified by
whether the focal firm is present as an incumbent
(set I) or is absent (complementary set 1*). Second,
markets can be classified by whether a given focalmarket rival is present (set J) or absent (complementary set J*). Finally, markets can be classified
into those that offer strong opportunities to share
resources with the focal market (set SR[m]) and
those that offer weak opportunities (complementary set SR*[m]). The independent variables were
calculated on the basis of the number of markets in
the intersections between these sets. Mathematical
FIGURE 2
Venn Diagram Illustrating the Construction of the Independent Variables
Universe of markets in the sample (N = 3,008)
I: Set of marketsin
_
which focal firm i is
present.
J: -Set of marketsin
which focal-market
rival\ j is present.
\*/
I*fnJnSR*[m]
Inj*nSR*[m]
\
InJnSR*[m]
,z'
SR[ml: Set of markets with strong ''
resource-sharing opportunities with
reore-hrn
focal market m. opotntiswt
\
t
~~~~~injnSR[m] /
\/
m_
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
Foca make
m
248
Academy of Management Journal
these two subsets correspond to the variables markets served (strong resource sharing) and markets
served (weak resource sharing), respectively. The
difference in the effects of these two variables reflects the impact of strong (versus weak) resourcesharing opportunities.
Multimarket contact. Following prior research,
we used a count measure of multimarket contact
(Baum & Korn, 1996; Evans & Kessides, 1994; Gimeno & Woo, 1996a). For a focal airline route, we first
counted the number of markets in which the airline
met a specific rival outside the focal market (that is,
the number of markets in subset InJ). Since a focal
market-unit can meet multiple focal-market rivals,
the variable multimarket contact was computed as
the average number of multimarket contacts with
all focal-market rivals. For instance, if a firm were
competing with two rivals in a focal market and
met the first in 100 other markets and the second in
300 markets, the measure of multimarket contact
would be 200 ([100 + 300]/2).8
Interaction effects. The common multiplicative
specification of interaction effects was not valid
here, since the independent variables for each focal
market-unit represented aggregates of the conditions in multiple nonfocal markets. It was possible
that a market-unit with large values for markets
served (strong resource sharing) and multimarket
contact might nevertheless have few multimarket
contacts in markets with strong resource-sharing
opportunities with the focal market. Instead, we
measured the interaction by splitting the previous
aggregate independent variable into submeasures
according to whether the moderating condition was
present or absent in each nonfocal market. Multimarket contact was split into two submeasures: (1)
multimarket contact (strong resource sharing),
which measured the extent of multimarket contact
in markets with strong resource-sharing opportunities with the focal market, and (2) multimarket
contact (weak resource sharing), which captured
multimarket contact in the remaining markets. The
first measure reflected the number of markets in
subset InJnSR[m], and the second equaled the
number of markets in subset IninSR* [ml.
Rivals' resource-sharing opportunities. To evaluate the effects of resource-sharing opportunities
available to rivals from markets not served by a
8 Since multimarket contact captures the markets of
overlap between firms, the same contacts are included in
the analysis of each firm's market-units. Although those
contacts are the same, they are used to predict dependent
variables for different market-units. This measurement
follows prior practice in research on contact.
June
focal firm, we divided the number of markets
served by focal-market rivals but not by the focal
firm into two subsets: those that had strong resource-sharing opportunities with the focal market
(subset I* nJnSR[mI) and those with weak resourcesharing opportunities (subset I*fnJnSR*[m]). The
variable rivals' noncontact markets served (strong
resource sharing) captured the number of markets
in the first subset, and the variable rivals' noncontact markets served (weak resource sharing)
captured the number in the second. The effect of
rivals' resource-sharing opportunities was evaluated from the difference in the effects of these two
measures.
Methodology and Control Variables
Since we used panel data composed of multiple
observations from each airline, market, and period,
there was potential for nonindependence of errors.
To accommodate the panel data structure, we used
the fixed-effects intercept model, also known as the
least squares dummy variable (LSDV) model
(Hsiao, 1986). Including fixed-effects intercepts for
each airline, city-pair market, and time period accounted for unobserved heterogeneity along those
dimensions. We also used dummy variables to account for the effects of mergers in the surviving
entities.
Three equations were estimated, each corresponding to one of the three dependent variablescost per revenue-passenger-mile, yield, and the
Lerner index. For each equation, we added a set of
variables to control for effects found to be significant in prior research (Baum & Korn, 1996; Borenstein, 1989; Evans & Kessides, 1994; Marin, 1995).
Control variables in the cost per revenue-passengermile equation included economies of scale in the
route (the number of passengers transported), the
cost of inputs, the percentage of direct flights, and
the airline's share of enplanements in the end-cities. Control variables in the yield equation included the average marginal cost among incumbents, demand characteristics, and other sources of
market power.9 Demand characteristics were reflected by market size, demand growth, the percentage of direct flights, the percentage of first-class
9 The need for those control variables can be established, for instance, from a Cournot oligopoly model
(Tirole, 1988) with heterogeneous costs, linear demand,
and conjectural variations (representing market power).
The equilibrium price in that model is a function of the
average marginal cost among incumbents, the intercept
of the demand function, the number of incumbents, and
conjectural variations.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
Gimeno and Woo
1999
tickets, and the percentage of round-trip tickets.
Other sources of market power included the number
of incumbents, the number of potential entrants,
the variance of the market shares of incumbents,
prior competitive experience among incumbents,
the airline's airport share, and the airline route's
share of the city-pair market. The control variables
in the profitability equation included the control
variables from the previous two equations. Appendix B explains the measurement of these control
variables.
Table 1 presents the descriptive statistics and
Pearson correlation matrix for all variables. We observed some substantial correlations among some
independent variables. This was not entirely surprising, since all the independent variables were
calculated from the number of nonfocal markets in
the six subsets in Figure 2. Although this situation
may raise concerns about multicollinearity, the theoretical and empirical validity of the study was not
compromised. Theoretically, the variables were
distinctly defined and often pertained to nonoverlapping sets of nonfocal markets. Even if the variables were correlated, it would be inappropriate to
assume that they were redundant. Empirically,
multicollinearity was reduced by the panel data
analysis method used.10 Variance inflation factors
for independent variables were never above 6,
below the rule of thumb of 10 often considered
to reflect excessive
multicollinearity
(Neter,
Wasserman, & Kutner, 1985). Even with moderate
multicollinearity, LSDV estimates are the best linear unbiased estimators (Greene, 1990). Since multicollinearity increases the estimated variance of
coefficients, our assessments of the significance of
individual coefficients are conservative.
RESULTS
Hypothesis 1 suggests that multimarket contact is
more likely to occur in markets with strong resource-sharing opportunities with a focal market.
We tested this prediction by comparing, for each
airline route, the proportions of multimarket contact with focal-market rivals in markets with strong
or weak resource-sharing opportunities with the
focal market. The proportion of multimarket contact in markets with strong resource-sharing opportunities was the ratio of multimarket contact
10 This is because the high zero-order correlations reflected cross-sectional differences in the number of markets served by different firms. These correlations were
reduced once airline-specific fixed effects were "partialed out."
249
(strong resource sharing) to markets served (strong
resource sharing), and the proportion of multimarket contact in markets with weak resource-sharing
opportunities was the ratio of multimarket contact
(weak resource sharing) to markets served (weak
resource sharing). Both proportions ranged from 0
to 1. For the sampled airline routes, the proportion
of multimarket contact in markets with strong resource-sharing opportunities averaged 0.5 (s.d. =
0.16). On average, a focal airline in a focal market
encounters its focal-market rivals in about half of
the nonfocal markets with strong resource-sharing
opportunities that it serves. The proportion of contacts in markets with weak resource-sharing opportunities averaged 0.38 (s.d. = 0.14). A paired t-test
comparison of these proportions yielded a difference in means of 0.13, significant at p < .001. Thus,
Hypothesis 1 was supported. The likelihood of an
airline's encountering a focal-market rival was
about 35 percent higher in markets that had strong
resource-sharing opportunities with the focal market. Hence, multimarket contact and resource-sharing opportunities were associated.
Hypothesis 2 explores the effect of resource-sharing opportunities with other markets on route
efficiency (see model 2 in Table 2). We observed
that the effect of markets served (strong resource
sharing) on costs per revenue-passenger-mile was
negative and significant (p < .001), but the coefficient was not significant for markets served (weak
resource sharing). The difference between these coefficients (b1 - b2), which reflects the differential
efficiency gained from a presence in markets with
strong resource-sharing opportunities with the focal market, is significant (p < .001). These findings led us to conclude that airlines were able to
obtain economies of scope from resource-sharing
opportunities across markets, in agreement with
Hypothesis 2.
Hypothesis 3a restates the widely tested relationship between multimarket contact and reduction in
rivalry. Results from the test of this hypothesis
(model 2 in Table 3) show that multimarket contact
had a significant and positive effect on yield and,
hence, a negative relationship to rivalry (p < .001).
Thus, our analysis supports Hypothesis 3a and is in
line with prior findings.
Hypothesis 3b states that the forbearance effect is
stronger when multimarket contact takes place in
markets with strong resource-sharing opportunities
with a focal market. Model 3 presents the results for
this hypothesis. In the analysis, we split multimarket contact into multimarket contact (strong resource sharing) and multimarket contact (weak resource sharing). The coefficient of the effect of
multimarket contact (strong resource sharing) on
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
z
z
zz
q
oz
61
81 LI
91 '51
Ne
)aPO
SIPAiX ,SIAI)
Xalu
S1ONJPU
asxnosa
0U
*?6tA'H
suo
PeAa
1eam)
j
uePip
'10d
>
go
}
Jo
STON STaNJPA
J0qIfl
J
910eluo9UOU
I
=
I
=
9
L
13lM U
V
?
* '
Buouw
1aNlJy\
jo O9uaOiJdxO0
s4t{i
aJPIs
aJPIS aJP1S
JO
J0Ua
od
0SPOAV
TJOdJ!V
oJ
OZIS
xOPUI
so,
,
jSP1U09J0d
IUM
u
aojpu-laSuossed
JeuiJPIfl
-aInUA
slasuassedjo
ODUBLUA
sluaqiun9ul
IelPuOd
sluaqiunrul
I
I
1 s)ndui
= =1
=
=
4so0
SSflD-Is.J
diil-punoljo
=
X
lf
P)lk
I
01fl}eJ
slue:
slu03 slu03slue:
junoD4unoDSOAAId
1$is
x
001
001IMOJD
TOJJIN
iod 000'1Ipaxenbs
dOd dd iod
paxenbs
0SMU03J10d
08PIU03a0d
0SP4U03J0d
08MPU03J0d
0PelU030d
0a1MU03J0d
001 a1
sioSuassied
'O)
allIu
allW
all0w alluw
UOIIIIlU
(10)(10) (1'0) (1'0)
sallui/paxenbs
(1
SJlajXeuu
S40)JefUl
SjlAJRLU
S4la0)1u1 S10)1J-uS10)PU-I
S1011'Tlu
001
001
001
001001-I 001
0
ZE
E
?zL
Sg
0 E'
? 8
s9aldo'I0
Z9
S10
6b-6 99g0
69@I16 9Z'0 1O'0
86WEZ1'? 89-117
V9-z
ZI0
00
0
Z
S
1b'b ZZo
09?
Z
suoielajioo
V
81
K1701-
ZO'?Z
90
60 ZZ
LO'ZZ' 00'1?-
00'
ZO-8L
80'?0'
10
91'- 60-91'- 60'- ?0- 76-
Zl
ZZ
OV-
?1T- 10
-
90
11T-
10T- 01
-
90-01'-
>
90-
*100-
80
80
10TZl180
-
10
S1l-
'Zl11
Z10-
?8-0 1708
Z?'11
8Z'
0
IZ'O Z1'0
b?'0Z
6L'
L0'
?1T- Zl
-
'zo
TO
Z8101'0
9?0 ?000
'
?0
d
TaNJPJA
s1uPJ1u0
J0qIfl
iqN
e
(Suixelns
I
=
8
uo9
Z104UO
4ixeluos
)eaM)
SUI1OJS)
SUO1S)
ioj
iai-
*1101 6
d
JO
4OPuIiIlIA
4aJPITJI1JAI
SIo3
paJOS oainose0
peA.OSpOAJOS
(SuiIrns
410elu
SleaXS
u1p
Z1
A
JOJ =
JJO9
VI ?1
01?1'- 1O ?S1Z''ZLI- ?1-LI80-b10'
11-?Z'- Z1--Z'Z
S1'0?' S'-
18-
60'8?-
10-
917
8LZ
Z' I1
ZO'- IZ-
80-I-
91'
L ?1
L' L-
IZ'
1 90'-
IZ-SZ'
9?'
80- ZO
1'
Z
9Z Z
1?
SZ ??-
VI II1
V
01
-
17 11
L'
81- 00
00'?0
?0- ZO'- 00'
SZ
bZ-- 91ZLO-
61 6Z'
9Z 60'
S1 19'
?'10'
LO-SO-
7l'-
10'-?17 ?0'ZZ-01-L-O'O'90- 90'
'Z-
1
17
Zl
SC?- 60
cZ 01
ZZ
-
tO0- 1S
Of 1S-
Zl
SZ
ZZL- 80-9Z- 81'-0?1-
10T-
17Z
IZ-
91-IZ-
11-
90' 81-b ZZ'
10
Sl
-
60-- ZO60-
10-
01- 11
61'1O7-?17-
90'-
?Z
11
-
91 11-
90'
L' ?1
81'
60'-
?0
-
91
?0-1'-
01-
01
Z0
-
1Z-
?Z- 11VT1-
171-
IS
-
OS
IZ ZS
IS-
8S
-
L
0F
ZL
Z9
Zl-
6?
-
66
?8
C1T-
9?
18
Zl1-
6?--
S1'
8?
?V OZ-SZ'
OZ'Z1
`Z
l
0? S0-
81--1?-
79'
c?
81- 1O'-Z
-'ZZ'-
1O'70-
10
80' 10-
ZZ-??
?0-C'-S0'
Z'-
IZ- Z1
9Z
OZ
90-?
ZZ'
S
'SuOl.eTAaQ
9
pue
-1bZ? ?
awaHV1
T
9
6
SUO9We10JJO9
O
JO
I1
IIV
ZI
E1
I
St
91
Li
91
61
OZ
ZZ
CZ
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
plpUPUlS
ZS
IZ
89
'SUOaW
t
01'
10-
??-?0'
0'
91-- LO'- ?17-O?'
??-
1O'9?' 6Z
00'
00'
81C
?0' 80
9S0o 61Z
171 ?1-
80'
0
L
?117
?1
IMA
1911 ueaw
Z8 99'91
s
p
?16
?88
q,esaOqeIlpA
Gimeno and Woo
1999
TABLE 2
Results of LSDV Regression Analysis: Effects on
Cost per Revenue-Passenger-Milea
Variable
Coefficient
Number of passengers
Direct flights
Cost of inputs
Airport share
Markets served (strong
resource sharing)
Markets served (weak
resource sharing)
N
Parameters estimated
R2
F
F for increment in R2
Degrees of freedom
Linear combinations
of coefficients:
b, - b2
Model 1
Model 2
-0.04***
(0.00)
4.09***
(0.10)
1.59***
(0.13)
-12.99***
(0.28)
-0.04***
(0.00)
3.97***
(0.10)
1.63***
(0.13)
-9.04***
(0.34)
-3.95***
(0.20)
-0.04
(0.03)
44,493
3,050
.79
51.42***
44,493
3,052
.79
52.02***
205.72***
2, 41,440
b1
b2
-3.91***
a Unstandardized regression coefficients are shown; standard
errors are in parentheses. Fixed effects for markets, airlines,
years, and mergers are not shown.
*** p < .001
yield is positive, large, and statistically significant
(p < .001). In contrast, the coefficient of multimarket contact (weak resource sharing) is not statistically significant in model 3 (it is marginally significant in model 4). The difference between these
coefficients is significant (p < .001). These results
support Hypothesis 3b and indicate that contacts in
markets with strong resource-sharing opportunities
with the focal market seem to have a greater forbearance effect than contacts in other markets.
Hypothesis 4 states that when rivals locate in
markets that provide strong resource-sharing opportunities and are not occupied by a focal firm,
rivalry for the focal market-unit will increase. In
model 4, we added two variables reflecting rivals'
positions in markets in which the focal firm was
absent, categorized by the strength of resourcesharing opportunities with the focal market. The
coefficient of rivals' noncontact markets served
(strong resource sharing) was negative, of large
magnitude, and significant (p < .001), and the coefficient of rivals' noncontact markets served (weak
resource sharing) was also negative and significant
(p < .01) but of substantially smaller magnitude.
The difference between these coefficients was statistically significant (p < .001). A focal market-unit
251
experienced more intense rivalry if focal-market
rivals were present in markets that offered them
scope economies that were not exploited by the
focal firm. Hypothesis 4 is supported.
Hypotheses 5 and 6a relate to the effect of focal
firm's resource-sharing opportunities and multimarket contact on profitability. Models 2 and 3 in Table
4 sequentially address these effects. Both effects
were statistically significant (p < .001) when introduced independently. Model 4 incorporates both
variables jointly. The results support Hypotheses 5
and 6a both and show that both multimarket
contact and resource-sharing opportunities have a
statistically significant and positive effect on
market-unit profitability. Although the coefficient
associated with multimarket contact (model 4) was
statistically significant (p < .001), it was less than
half of the magnitude of the corresponding coefficient in model 3, an analysis in which resourcesharing opportunities were not controlled for.
Thus, the effect of multimarket contact on performance can be overestimated (in this case, by a
factor of over 100 percent) if the effects of resourcesharing opportunities are not controlled for. The
coefficients in model 5 also suggest that a presence
in markets with strong resource-sharing opportunities with the focal market has a more positive impact on profitability than a presence in markets
with weak resource-sharing opportunities, since
the coefficients of these variables were significantly
different (p < .001). Thus, Hypotheses 5 and 6a are
supported."
Hypothesis 6b explores whether the effect of
multimarket contact on profitability depends on
the strength of resource-sharing opportunities with
a focal market. As for the tests of Hypothesis 3b, we
divided multimarket contact into two submeasures
based on whether contacts occurred in markets
with strong or weak resource-sharing opportunities
with the focal market. The coefficient of multimarket contact (strong resource sharing) was positive,
of large magnitude, and statistically significant
(p < .001), but the coefficient of multimarket contact (weak resource sharing) was negative, of small
" It is interesting to note that although markets served
(weak resource sharing) does not have a significant effect
on efficiency, the variable has a positive and significant
effect on profitability. The combination of those effects
may point to the presence of revenue superadditivity if
customers in one market have a preference for firms that
are present in other markets. This may be due to customers' perceiving multimarket firms as reputable or reliable.
When airlines offer frequent-flier programs, customers
may also prefer to accumulate miles in airlines that have
more extensive networks.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
June
Academy of Management Journal
252
TABLE 3
Results of LSDV Regression Analysis: Effects on Yielda
Model
Variable
Coefficient
Average marginal cost among
incumbents
Market size
Market growth
Percentage of round-trip tickets
1
2
3
4
0.79***
(0.04)
0.30***
0.80***
(0.04)
0.31***
0.81***
(0.04)
0.31***
0.82***
(0.04)
0.31***
(0.01)
(0.01)
(0.01)
(0.01)
0.01
0.01
0.01
0.01
(0.01)
(0.01)
(0.01)
(0.01)
-5.54***
(0.18)
21.94***
(0.62)
0.12
(0.08)
7.25***
(0.25)
0.00
Percentage of first-class tickets
Direct flights
Airport share
Prior competitive experience
(0.00)
-0.38***
(0.02)
-0.54***
(0.03)
-0.83***
(0.24)
0.57***
(0.12)
Number of potential entrants
Number of incumbents
Market share variance
Market share
Multimarket contact
Multimarket contact (strong
resource sharing)
Multimarket contact (weak
resource sharing)
Rivals' noncontact markets served
(strong resource sharing)
Rivals' noncontact markets served
(weak resource sharing)
N
Parameters estimated
R2
F
F for increment in R2 b
Degrees of freedom
-5.56***
(0.18)
21.86***
(0.62)
0.19*
(0.08)
7.17***
(0.25)
0.00
(0.00)
-0.36***
(0.02)
-0.50***
(0.03)
-0.76**
(0.24)
0.51***
(0.12)
0.16***
(0.02)
-5.61***-5.62**
(0.18)
21.75***
(0.62)
0.20*
(0.08)
6.87***
(0.25)
0.00
(0.00)
-0.35***
(0.02)
-0.46***
(0.03)
-0.65**
(0.24)
0.47***
(0.12)
2.31***
(0.32)
0.03
(0.03)
b
b2
(0.18)
21.70***
(0.61)
0.27***
(0.08)
6.46***
(0.27)
0.00
(0.00)
-0.37
(0.02)
-0.51*
(0.03)
-0.61*
(0.24)
0.37**
(0.12)
1.75***
(0.37)
0.05k
(0.03)
b3
-0.97
b4
(0.25)
-0.04**
(0.01)
44,493
3,058
.89
107.45***
44,493
3,059
.89
107.65***
83.79***
1, 41,433
Linear combinations of
coefficients:
b, - b2
b3 - b4
44,493
3,060
.89
107.74***
44.16***
1, 41,432
2.28***
44,493
3,062
.89
107.82***
25.64***
2, 41,430
1.69***
-0.93
a
Unstandardized regression coefficients are shown; standard errors are in parentheses. Fixed effects for markets, airlines, years, and
mergers are not shown.
b
Fs are for model 2 versus model 3, model 3 versus model 2, and model 4 versus model 3, respectively.
tp
*p
**
***
< .10
< .05
p < .01
p < .001
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1999
Gimeno and Woo
253
TABLE 4
Results of LSDV Regression Analysis: Effects on Lerner Indexa
Model
Variable
Coefficient
Average marginal cost among
incumbents
Market size
Market growth
Percentage of round-trip tickets
Percentage of first-class tickets
Direct flights
Airport share
Prior competitive experience
1
Market share variance
Market share
Number of passengers
b,
Multimarket contact (strong
resource sharing)
Multimarket contact (weak
resource sharing)
Rivals' noncontact markets served
(strong resource sharing)
Rivals' noncontact markets served
(weak resource sharing)
b4
N
Parameters estimated
R2
F
F for increment in R21'
Degrees of freedom
Linear combinations of
coefficients:
b, - b2
b - b
b6 - b7
6
1.38***
(0.29)
0.11
1.38***
(0.29)
0.07
1.43***
(0.29)
0.12
(0.09)
(0.09)
(0.09)
(0.09)
(0.09)
(0.09)
0.18**
(0.06)
-15.13***
(1.26)
117.83***
(4.26)
-18.50***
(0.58)
81.91***
(1.70)
0.03**
0.20***
(0.06)
-16.66***
(1.24)
112.65***
(4.22)
-17.19***
(0.57)
55.29***
(1.96)
0.04***
0.19***
(0.06)
-15.31***
(1.25)
117.00***
(4.25)
-17.82***
(0.58)
81.11***
(1.70)
-0.01
0.20***
(0.06)
-16.70***
(1.24)
112.51***
(4.22)
-16.92***
(0.58)
55.80***
(1.96)
0.02*
0.20***
(0.06)
-16.84***
(1.24)
112.28***
(4.21)
-17.00***
(0.58)
58.36***
(1.99)
0.02*
0.18**
(0.06)
-16.82***
(1.24)
112.36***
(4.21)
-16.47***
(0.58)
58.36***
(1.99)
0.04***
-6.87***
(0.74)
Markets served (strong resource
sharing)
Markets served (weak resource
sharing)
Multimarket contact
5
0.98***
(0.30)
0.27**
(0.01)
Cost of inputs
4
1.35***
(0.29)
0.06
-1.53***
(0.14)
-3.07***
(0.20)
-2.41
(1.65)
10.23***
(0.82)
0.15*
Number of incumbents
3
0.86**
(0.30)
0.18'
(0.01)
Number of potential entrants
2
b2
(0.01)
-1.38***
(0.13)
-2.70***
(0.19)
-1.65
(1.63)
8.76***
(0.82)
0.14***
(0.01)
-7.88***
(0.73)
28.62***
(1.08)
1.56***
(0.14)
b3
(0.01)
-1.36***
(0.14)
-2.68***
(0.20)
-1.69
(1.65)
9.59***
(0.82)
0.15k*
(0.01)
-7.01***
(0.74)
1.61***
(0.12)
(0.01)
-1.32***
(0.13)
-2.54***
(0.20)
-1.31
(1.63)
8.46***
(0.82)
0.14***
(0.01)
-7.87***
(0.73)
27.70***
(1.09)
1.35***
(0.14)
0.77***
(0.13)
b5
(0.01)
-1.21***
(0.14)
-2.14***
(0.20)
-0.46
(1.64)
8.54***
(0.82)
0.14***
(0.01)
-8.04***
(0.73)
14.66***
(1.76)
1.60* * *
(0.16)
21.67* * *
(2.79)
-0.44*
(0.21)
21.50* * *
(2.80)
-0.48*
(0.21)
-13.33***
(2.23)
-0.08
(0.12)
b7
44,493
3,062
.56
16.96***
463.42***
2, 41,430
44,493
3,061
.55
16.43***
174.45***
1, 41,431
27.06***
(0.01)
-7.87***
(0.73)
21.65***
(1.36)
1.60* * *
(0.15)
b6
44,493
3,060
.55
16.31***
(0.01)
-1.35***
(0.14)
-2.54***
(0.21)
0.08
(1.64)
7.48***
(0.83)
0.14***
44,493
3,063
.56
16.99***
37.38***
1, 41,429
44,493
3,064
.56
17.02***
56.15***
1, 41,428
44,493
3,066
.56
17.06***
37.50***
2, 41,426
26.35***
20.06***
22.12***
13.06***
21.97***
-13.25***
a
Unstandardized regression coefficients are shown; standard errors are in parentheses. Fixed effects for markets, airlines, years, and
mergers are not shown.
b
Fs are for models 2 and 3 versus model 1, model 4 versus model 2, model 5 versus model 4, and model 6 versus model 5, respectively.
tp
p
** p
*** p
*
<
<
<
<
.10
.O5
.01
.001
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
254
Academy of Management Journal
magnitude, and less significant (p < .05). The difference between these coefficients was highly statistically significant (p < .001). The finding suggests that multimarket contact has the greatest
effect on profitability when it occurs in markets
that have strong resource-sharing opportunities
with a focal market. Thus, Hypothesis 6b is supported.
Hypothesis 7 suggests that the profitability of a
focal market-unit will be lower when the focal firm
is not present in markets occupied by focal-market
rivals and these markets provide the latter with
strong resource-sharing opportunities. As for Hypothesis 4, we segmented the markets in which the
focal-market rivals were present but the focal firm
was absent by the strength of resource-sharing opportunities. The coefficient of rivals' noncontact
markets served (strong resource sharing) was negative, of large magnitude, and significant (p < .001),
although the coefficient of rivals' noncontact markets served (weak resource sharing) was statistically insignificant. The difference between these
coefficients was statistically significant (p < .001).
The profitability of a market-unit decreases if focalmarket rivals are present in markets that give those
rivals some economies of scope that are not exploited by a focal firm. Hypothesis 7 is supported.
Overall, the results for profitability (Hypotheses 5
to 7) are very consistent with the findings for efficiency (Hypothesis 2) and intensity of rivalry (Hypotheses 3 and 4).
DISCUSSION
This article is the first to systematically investigate the association between multimarket contact
and the presence of resource-sharing opportunities
(the antecedent of economies of scope) and their
joint effects on efficiency, rivalry, and, ultimately,
profitability. We found that multimarket contacts
were more likely to occur in markets characterized
by strong resource-sharing opportunities with a focal market. Market-units within a firm that share
common resources are also likely to encounter
common competitors. This possibility has been recognized in the literature (Porter, 1985), but we have
provided the first explicit empirical evidence supporting it. This finding suggests that the effects of
economies of scope and multipoint competition
may be entangled and that they should be simultaneously considered in future studies as two interrelated dimensions.
The empirical model provides evidence linking
scope economies and multimarket contact with
profitability. We have also provided some evidence
about the causal mechanisms linking the indepen-
June
dent variables to profitability by examining their
links to two important antecedents of profitability:
efficiency and intensity of rivalry. The results with
all three dependent variables were internally consistent and in full support of theoretical predictions. With respect to scope economies, we found
that the efficiency of a market-unit was enhanced,
and its profitability was accordingly increased, if
the focal firm participated in markets that had
strong resource-sharing opportunities with the focal market. Multimarket contact was found to reduce the intensity of the rivalry experienced by a
market-unit and, accordingly, to increase its profitability. Moreover, the effects of multimarket contact on the intensity of rivalry and profitability
were more pronounced if the contacts occurred in
markets with strong resource-sharing opportunities
with the focal market. This latter finding suggests a
boundary condition to the forbearance effect of
multimarket contact. It appears that multimarket
contact has little effect in the absence of strong
economies of scope. Finally, we found that intensity of rivalry was increased, and profitability accordingly decreased, if a focal firm was absent from
markets where focal-market rivals obtained strong
economies of scope. Without the competitive restraint enforced by multimarket contact, the superior efficiency of rivals can make them more capable and aggressive competitors, increasing the
intensity of rivalry for the focal market-unit and
reducing its profitability.
In addition to examining the significance of the
profitability effects of economies of scope and multimarket contact, we also examined their magnitude. To that effect, consider a representative market-unit obtaining a Lerner index of 11.61, the
average in the sample. We investigated the change
in the Lerner index predicted by a change in the
number of nonfocal markets in each of the six subsets determined by the intersection of the I, J, and
SR[m] sets. An arbitrary number of 25 markets was
chosen; this number was less than two standard
deviations of the number of markets in each subset.
Thus, a change of 25 markets within a subset was a
realistic change in this sample. If a focal firm added
25 markets that had strong resource-sharing opportunities with the focal market and were not served
by the focal-market rivals, the profitability of the
focal market-unit would increase by 31.56 percent.
However, if focal-market rivals added 25 markets
with strong resource-sharing opportunities and not
served by the focal firm, the latter's profitability
would decrease by 28.70 percent. Taken together,
these findings suggest that when both a focal firm
and its focal-market rivals obtain economies of
scope from different nonoverlapping markets, the
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
Gimeno and Woo
1999
profitability of the market-unit is roughly the same
as if neither firm had obtained scope economies. In
that case, the profitability effect of economies of
scope is "competed away" by equally efficient and
unrestrained rivals. In contrast, the profitability of
a market-unit increases by 77.86 percent if the focal
firm adds 25 markets with strong resource-sharing
opportunities with the focal market and providing
multimarket contact with the focal-market rivals.
Hence, the results suggest that the best profitability
scenario for a focal market-unit is obtained when
presence in nonfocal markets simultaneously provides economies of scope and multimarket contact
with rivals. The worst profitability scenario for a
focal market-unit occurs when it lacks scope economies while competing against rivals with strong
scope economies.
If a focal firm adds 25 markets with weak resource-sharing opportunities with the focal market
and in which the focal-market rivals are not
present, profitability is 3.45 percent higher. The
increase is only 2.41 percent if these markets provide multimarket contact, suggesting that multimarket contact does not contribute to profitability
in markets with weak resource-sharing opportunities. When the focal-market rivals add 25 markets
with weak resource-sharing opportunities and not
served by the focal firm, profitability of the focal
market-unit decreases by only 0.17 percent. Overall, the profitability effects of presence in markets
with weak resource-sharing opportunities are of
much smaller magnitude than the effects of presence in markets with strong resource-sharing opportunities.
Generalizability
of Results
This study benefited from the rich data available
in the airline industry, which allowed the measurement of independent and dependent variables at
very disaggregate levels. Moreover, the industry
was possibly an ideal context for finding support
for the theoretical predictions. In the airline industry, it appears that economies of scope are strong
and well understood by managers. The forbearance
effects associated with multimarket contact are also
of important magnitude in the industry, possibly
because of the high concentration in most city-pair
markets. The combination of these dimensions
leads to our conclusion that airlines are better off
pursuing economies of scope in the same markets
as competitors, since the forbearance effect associated with multimarket contact reduces the erosion
of performance. The generalizability of these results needs to be evaluated in other settings (for
instance, multiproduct industries or international
255
and corporate diversification), since the relative
strength of scope economies and multimarket contact may differ across settings. In other settings,
scope economies may be of a less technical nature,
and there may be differences in firms' ability to
implement those economies. Forbearance may also
be less viable in other settings if markets are defined in a more aggregate way and include more
competitors. In any case, the generalizability of our
results to other settings is an empirical question
that must be established by future research.
Implications for Multipoint Competition
Research
The theory and results presented in this article
have important implications for the literature on
multipoint competition because they allow a more
detailed understanding of the role of forbearance in
competitive situations. Bernheim and Whinston's
(1990) very influential game-theoretical treatise on
multipoint competition developed a model of mutual forbearance that, as a simplifying assumption,
did not incorporate the effect of economies of scope
among the markets in which contact occurred. We
show that this assumption can lead to a serious
misrepresentation of the actual conditions under
which multimarket contact occurs and can also
lead to overestimation of the effect of multimarket
contact. Indeed, we found that the forbearance effect of multimarket contact is less likely to exist if
the multiple markets are not linked by scope economies.
The interaction between multimarket contact
and scope economies has important theoretical implications. Mutual forbearance appears to be a
mechanism by which multipoint competitors retain the value created by their scope economies, by
avoiding its loss through price competition. Their
ability to do so presumes cost advantages from
synergistic cross-unit activities that are not available to single-point incumbents and potential entrants. Thus, when competitors meet in multiple
markets with strong scope economies, they may
maintain prices above their own efficient costs (given their own scope economies) but below the limit
price that would encourage entry or challenges by
less efficient firms. Without multimarket contact,
these same efficient competitors would reduce
their prices toward their cost levels, eroding their
efficiency quasi-rents. Ironically, the interaction
also suggests that mutual forbearance behavior
would be quite unlikely among diversified firms of
the conglomerate form, the context in which the
mutual forbearance hypothesis was originally developed. Multipoint competition theory may be
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
256
Academy of Management Journal
most relevant under conditions of strong resource
sharing among market-units, such as those existing
in the airline, packaged foods, and telecommunications industries and under conditions of related
diversification or geographic expansion.
Finally, in agreement with Montgomery (1994),
who was quoted in this article's introduction, we
caution against the possible omitted variable bias
in studies of mutual forbearance that do not control
for the effects of economies of scope. Since multimarket contact is more likely to occur in markets
with strong scope economies, the variable may pick
up the effect of those omitted economies unless
they are explicitly controlled for. Our analysis
showed that lack of control for resource-sharing
opportunities may lead to overestimation of the
effect of multimarket contact on performance by
over 100 percent.
Implications for Economies of Scope Research
This study also has important implications for
research on the performance effect of scope economies. In prior research examining this effect (for
instance, in the corporate and international diversification literature), the role of rivals' productmarket scope has not been considered. According
to our model, this is problematic. In our model,
scope economies make market-units more efficient
and, accordingly, more profitable. However, profitability not only depends on the efficiency of a focal
market-unit, but also on the efficiency of its rivals
and on the aggressive or collusive interaction with
those rivals. A market-unit with strong scope economies but competing with rivals that also have
strong scope economies (from nonoverlapping markets) and aggressive competitive behavior (due to
lack of forbearance) obtains the same profitability
as a market-unit without scope economies competing with rivals that also lack scope economies. This
result suggests that market-units with scope economies will perform better than those without them
under two alternative conditions: (1) when they
compete with rivals lacking scope economies or (2)
when they compete with rivals that have similar
scope economies but are less aggressive because of
multimarket contact. These important considerations need to be taken into consideration in future
research on the performance effect of scope economies.
Extensions
Our focus on the interaction between multimarket contact and scope economies has left some relevant dimensions unattended. Firms with overlap-
June
ping multimarket scope may also encounter each
other as buyers or suppliers or as partners in cooperative activities. For example, Disney and Time
Warner, which compete in multiple markets
(amusement parks, specialty stores, TV broadcasting, video), also encounter each other as partners in
several cooperative ventures and as buyers or suppliers to each other (Landro, Reilly, & Turner,
1993). This situation makes competitive outbreaks
between these companies particularly complex,
since firms may react on any of the dimensions of
interaction. We did not consider other possible vertical or cooperative links among the firms in this
study, but this is an important direction for extending current research (Gimeno & Woo, 1996b).
We also encourage extension of this research to
the contexts of international or corporate diversification in which different organizational units
(country subsidiaries, divisions) are responsible for
operations in different markets. This line of research would explicitly highlight the role of organizational structure and coordinating mechanisms
on the effects of resource-sharing opportunities and
multimarket contact. We argued that lack of administrative coordination may limit the ability of firms
to obtain the forbearance effect of multimarket contact and assumed that strong resource-sharing opportunities between markets were likely to be associated with the presence of coordinated decision
making across markets. Explicit observation of administrative coordinating mechanisms across markets would make a substantial contribution by allowing tests of their moderating influence on the
efficiency effects of resource-sharing opportunities
and the forbearance effects of multimarket contact.
The role of organizational structure must be integrated more fully into the study of multipoint competition, since firms differ in their orientations toward internal competition or cooperation (Collis &
Montgomery, 1997; Hill, Hitt, & Hoskisson, 1992).
Studies of multipoint competition in the global
context could benefit from a direct examination of
the coordinating mechanisms used among national
subsidiaries.
Conclusion
This article has highlighted the interdependence
between economies of scope and multipoint competition. Market-units within a firm that share common resources are also likely to encounter common
competitors. Multipoint competition is likely to become even more prevalent as firms expand to take
advantage of resource-sharing opportunities, as the
boundaries between markets become blurred, and
as firms become global competitors. Given the sig-
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1999
Gimeno and Woo
nificant interaction between multimarket contact
and scope economies, future research on scope
economies and multimarket contact should address
them jointly. In particular, the competitive context
of multimarket firms and the product-market scope
of rivals should be explicitly considered in the
well-established research on the performance effects of scope economies. Further cross-fertilization
between research on scope economies and multipoint competition is needed.
REFERENCES
Ansoff, H. I. 1965. Corporate
McGraw-Hill.
strategy.
New York:
Areeda, P., & Turner, D. 1979. Conglomerate mergers:
Extended interdependence and effects on interindustry competition as grounds for condemnation.
University of Pennsylvania
Law Review, 127:
1082-1103.
Bailey, E. E., & Friedlaender, A. F. 1982. Market structure
and multiproduct industries. Journal of Economic
Literature, 20: 1024-1048.
257
rivalry: Toward a theoretical integration. Academy
of Management Review, 21: 100-134.
Collis, D. J., & Montgomery, C. A. 1997. Corporate strategy: Resources and the scope of the firm. Chicago:
Irwin.
Cool, K., Dierickx, I., & Jemison, D. 1989. Business strategy, market structure and risk-return relationships:
A structural approach. Strategic Management journal, 10: 507-522.
Davis, R., & Thomas, L. G. 1993. Direct estimation of
synergy: A new approach to the diversity-performance
debate.
Management
Science,
39:
1334-1346.
Demsetz, H. 1973. Industry structure, market rivalry and
public policy. Journal of Law and Economics, 16:
1-9.
Edwards, C. D. 1955. Conglomerate bigness as a source of
power. In Business concentration and price policy
(National Bureau of Economic Research conference
report): 331-352. Princeton, NJ: Princeton University Press.
Bailey, E. E., Graham, D. R., & Kaplan, D. P. 1985. Deregulating the airlines. Cambridge, MA: MIT Press.
Evans, W. N., & Kessides, I. N. 1994. Living by the "golden rule." Multimarket contact in the US airline industry. Quarterly Journal of Economics, 109: 341366.
Barnett, W. P. 1993. Strategic deterrence among multipoint competitors. Industrial
and Corporate
Change, 2: 249-278.
Feinberg, R. M. 1984. Mutual forbearance as an extension
of oligopoly theory. journal of Economics and Business, 36: 243-249.
Baum, J. A. C., & Korn, H. J. 1996. Competitive dynamics
of interfirm rivalry. Academy of Management lournal, 39: 255-291.
Gimeno, J. 1999. Reciprocal threats in multimarket rivalry: Staking out "spheres of influence" in the U.S.
airline industry. Strategic Management Journal,
20: 101-128.
Bernheim, D., & Whinston, M. D. 1990. Multimarket contact and collusive behavior. RAND Journal of Economics, 21: 1-26.
Berry, S. T. 1989. Entry in the airline industry. Unpublished doctoral dissertation, University of Wisconsin, Madison.
Boeker, W., Goodstein, J., Stephan, J., & Murmann, J. P.
1997. Competition in a multimarket environment:
The case of market exit. Organization Science, 8:
126-142.
Borenstein, S. 1989. Hubs and high fares: Dominance and
market power in the US airline industry. RAND
Journal of Economics, 20: 344-365.
Brander, J. A., & Zhang, A. 1990. Market conduct in the
airline industry: An empirical investigation. RAND
Journal of Economics, 21: 567-583.
Brueckner, J. K., Dyer, N. J., & Spiller, P. T. 1992. Fare
determination in airline hub-and-spoke networks.
RAND Journal of Economics, 23: 309-333.
Brush, T. H. 1996. Predicted change in operational synergy and post-acquisition performance of acquired
businesses. Strategic Management Journal, 17:
1-24.
Chen, M. 1996. Competitor analysis
and inter-firm
Gimeno, J., & Woo, C. 1996a. Hypercompetition in a
multimarket environment: The role of strategic similarity and multimarket contact in competitive deescalation. Organization Science, 7: 322-340.
Gimeno, J., & Woo, C. 1996b. Economic multiplexity: The
structural embeddedness of cooperation in multiple
relations of interdependence. In J. A. C. Baum & J.
Dutton (Eds.), Advances in strategic management,
vol. 13: 323-361. Greenwich, CT: JAI Press.
Greene, W. H. 1990. Econometric
Macmillan.
analysis. New York:
Hill, C. W. L., Hitt, M. A., & Hoskisson, R. E. 1992.
Cooperative versus competitive structures in related
and unrelated diversified firms. Organization Science, 3: 501-521.
Hsiao, C. 1986. Analysis of panel data. New York: Cambridge University Press.
Jayachandran, S., Gimeno, J., & Varadarajan, P. R. 1999.
Multimarket competition and interfirm rivalry: A
synthesis and implications for marketing strategy.
Journal of Marketing, 63: In press.
Karnani, A., & Wernerfelt, B. 1985. Multiple point competition. Strategic Management Journal, 6: 8 7-96.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
258
Academy of Management Journal
Landro, L., Reilly, P. M., & Turner, R. 1993. Clash of
titans. Wall Street journal, April 14: Al.
Marin, P. L. 1995. Competition in European aviation:
Pricing policy and market structure. Journal of Industrial Economics, 63: 141-159.
McWilliams, A., & Smart, D. L. 1993. Efficiency v. structure-conduct-performance: Implications for strategy
research and practice. Journal of Management, 19:
63-78.
June
the enterprise. Journal of Economic Behavior and
Organization, 1: 223-245.
Tirole, J. 1988. The theory of industrial
Cambridge, MA: MIT Press.
Young, G., Smith, K. G., & Grimm, C. M. 1997. Multimarket contact, resource heterogeneity, and rivalrous firm behavior. Academy of Management Best
Paper Proceedings: 55-59.
Montgomery, C. A. 1994. Corporate diversification. journal of Economic Perspectives, 8: 163-178.
Montgomery, C. A., & Hariharan, S. 1991. Diversified
expansion by large established firms. journal of Economic Behavior and Organization, 15: 71-89.
Neter, J., Wasserman, W., & Kutner, M. H. 1985. Applied
linear statistical models. Homewood, IL: Irwin.
Nomani, A. Q. 1990. Fare game: Airlines may be using a
price-data network to lessen competition. Wall
Street journal, June 28: A1-A8.
O'Brian, B. 1994. Southwest Air says it will add longer
flights. Wall Street Journal, July 5: A3.
Panzar, J. C., & Willig, R. D. 1981. Economies of scope.
American Economic Review, 71: 268-272.
Parker, P. M., & R6ller, L. H. 1997. Collusive conduct in
duopolies: Multimarket contact and cross-ownership in the mobile telephone industry. RAND journal of Economics, 28: 304-322.
APPENDIX A
Mathematical Definition of Independent Variables
We first define two matrixes of dummy variables, A
and SR. A is a three-dimensional matrix whose element
Aimt equals 1 if firm i is present in marketm at time t and
equals 0 otherwise. SR is a two-dimensional square matrix whose element SRin7 equals 1 if market n has resource-sharingopportunities with marketm and 0 otherwise. Given those two matrixes, the independent
variables reflecting resource-sharing opportunities for
the focal market-unit im at time t are operationally defined as follows:
Markets served (strong resource sharing)imt
Porter, M. E. 1980. Competitive strategy. New York: Free
Press.
Porter, M. E. 1985. Competitive advantage.
Free Press.
New York:
Schmalensee, R. 1987. Collusion versus differential efficiency: Testing alternative hypotheses. Journal of
Industrial Economics, 35: 399-425.
Scott, J. T. 1993. Purposive diversification and economic performance. Cambridge, UK: Cambridge
University Press.
Smith, F. L., & Wilson, R. L. 1995. The predicted validity
of the Karnani and Wernerfelt model of multipoint
competition. Strategic Management Journal, 16:
143-160.
Teece, D.
J. 1980. Economies of scope and the scope of
Aint
X SRmn
nUm
Markets served (weak resource sharing)in,t
Aint X (1 -
E
SRmn).
nfm
The next independent variables are measured relative to
focal market rivals. We first calculate the variables with
respect to a given focal-marketrival j (in brackets) and
then obtain the average of the measures for all focalmarket rivals. Thus:
Multimarket contactin,t
Scheffman, D. T., & Spiller, P. T. 1996. Econometric
market delineation. Managerial and Decision Economics, 17: 165-178.
Scherer, F. M., & Ross, D. 1990. Industrial market structure and economic performance. Boston: Houghton
Mifflin.
E
-
Penrose, E. 1959. The theory of the growth of the firm.
Oxford: Oxford University Press.
Peteraf, M. A. 1993. The cornerstone of competitive advantage: A resource-based view. Strategic Management journal, 14: 179-191.
organization.
=
X
Ajmt
[
[ nni
j&i
A
X
x Ajilt
1
Ajnt
Multimarket contact (strong resource sharing)in,t
Ajmt
[
Aint
X
Ajnt X SRmn
I nfm
Z
J
i
Ajmt.
Multimarket contact (weak resource sharing)int,
=EAjmt
&i
X
Aint X
Ajilt x (
-SRmn)
1
Ajmt.
j
n:nm
i
Rivals' noncontact markets served
(strong resource sharing)im7t
=EAjmt
&i
X [
t) x Ajnt x SR1nn
nm
n
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions
1
j
Ajmt.
i
1999
Gimeno and Woo
Rivals' noncontact markets served
(weak resource sharing)imt
=Ajmt X[(i
-
-
Aint) X Ajnt X (1-
SRmn)1/Ajmt
nfm
j#i
APPENDIX B
Definition of Control Variables
Average marginal cost among incumbentsmt: The average
cost per available-seat-mileimt among incumbents in the
market. Cost per available-seat-mile was used (instead of
cost per revenue-passenger-mile) because it better reflects the marginal (as opposed to average) costs, which
are the determinants of pricing strategies.
Market sizemt: Market gravity, or the product of the total
constant-dollar personal income in the end-cities' Metropolitan Statistical Areas divided by the square of the
distance between the cities (data were from the Regional
Economic Information System).
Market growthmt: The growth in market gravity from a
prior to a current year (representing growth in population
or per capita income in the end-cities).
Percentage of round-trip ticketsimt: The percentage of an
airline route's passengers buying round-trip tickets.
Percentage of first-class ticketsimt: The percentage of an
airline route's passengers flying first class.
Direct flightsjmt: The percentage of an airline route's passengers flying nonstop.
Airport shareimt: The average of a firm's share of total
enplanements at both end-cities of a city-pair market.
Prior competitive
experiencemt:
The minimum
age of an
incumbent in a market. The age of the youngest rival
limits prior competitive experience among incumbents.
Number of potential entrantsmt: The number of firms
with presence at both end-cities of a city-pair market that
are not incumbent in the market.
259
Number of incumbentsmt: The number of firms that are
incumbent in the market (that either have an over 5
percent market share or carry at least 900 passengers per
quarter).
Market shareimt: The percentage of total passengers transported in the market carried by the focal airline route.
Market share variancemt: The variance of market shares
among incumbents.
The number of passengers carNumber of passengersimt:
ried by the focal airline route.
Cost of inputsmt: An index of changes in the cost of labor
and fuel costs, measured as the Standard Industry Fare
Level (an index calculated by the FAA to estimate "fair"
prices in each market and adjusted by changes in costs of
inputs) per mile of distance.
Merger controlsit:
Dummy variables to account for the
changes to ongoing companies from major mergers and
acquisitions. For the entity that survived a merger, a
dummy variable was set to 1 for periods after the merger
and to 0 for periods before the merger.
Javier Gimeno (Ph.D., Purdue University) is an assistant
professor of management at Texas A&M University. His
current research interests include multipoint competition, competitive strategy in global and domestic settings, and the dynamics of competitive and cooperative
interactions. He is also interested in the survival and
performance of new entrepreneurial firms.
Carolyn Y. Woo (Ph.D., Purdue University) is the Raymond and Milann Siegfried Chair in Entrepreneurial
Studies and the Martin G. Gillen Dean of the College of
Business Administration at the University of Notre
Dame. Her teaching and research interests include corporate and competitive strategy analyses, manufacturing
strategy, entrepreneurship, management of innovation
and change, enterprise integration, and organizational
systems.
This content downloaded on Wed, 6 Feb 2013 05:54:33 AM
All use subject to JSTOR Terms and Conditions